1. Norms in Small-Number Relationships

Small-number norms are relatively easily explained from rational-choice principles. Consider one small-number norm: promise keeping. Bicchieri discusses this norm more or less only in passing and assumes its force in many instances of supporting or complying with other norms. One might find this surprising since the norm of promise keeping is possibly the most commonly cited norm in all of the literature, especially the literature of twentieth-century philosophy. But neglecting this norm is not surprising in an important way because it is hard to make sense of such a norm in one-off relationships, which is the modal structure of the gaming literature. One might suppose that any theory of norms that cannot handle promise keeping is a nonstarter, or one could insist that promise keeping is somehow embedded in many of the cooperation games that generate claims about norms. I think the latter claim is likely true, but if so, then we should want to see a direct effort by the gamers to explain promise keeping.

There is a class of important universalistic norms that is virtually self-enforcing in many contexts. These are norms such as those for telling the truth, keeping promises, and maintaining fidelity to spouses and friends. They are self-enforcing when they govern ongoing relationships between pairs or very small numbers of people. In brief, I will refer to these as dyadic norms. The enforcement of these norms comes naturally from the fact that the relationships that they govern are of value to the participants beyond the instant interaction on which someone must keep the relevant norm or violate it. These characteristics of the relationship suggest that it is an iterated prisoner's dilemma in its incentive structure. Each participant in the dyad sometimes has a short-term interest in violating the norm but a long-term interest in maintaining the relationship. If the latter is great enough, it can trump the short-term interest and make it worthwhile to forgo short-term gains in the interest of longer term gains (see Hardin Reference Hardin1988, Chapter 2).

One might tell the truth out of strict moral scruples, but one also has an interest in telling the truth to one's ongoing relations. Even if you are bound by moral scruples, you may nevertheless depend on the incentives of self-interest that keep others honest enough to make dealing with them worth your while. The usual resolution of an iterated prisoner's dilemma in which one party fails to cooperate often enough is not to join that party in cheating but simply to withdraw from the relationship. That can be a brutal sanction. If others have an interest in cooperation with you, you have better prospects from cooperating than from withdrawing or cheating. In general, you have better life prospects if those around you recognize their interest in maintaining cooperative relations with you (Hardin Reference Hardin2002, Chapter 1). Being cooperative loses much of its value if too few others are cooperative, and it may finally even become disvalued. Similarly, truth telling may often seem to serve perverse purposes for many relationships in a Nazi or other totalitarian society. Typically, then, these dyadic norms are straightforwardly self-enforcing and not inherently moral. Such a norm may govern a very large population, all of whom are involved with various others in dyadic relationships. But there is no large-number equivalent of the dyadic norms. Many, perhaps most, the game experiments focus on dyads.

2. Norms in Large-Number Relationships

A norm that governs relationships that are inherently large-number rather than dyadic cannot be reinforced merely by the withdrawal of particular cooperative parties from interactions with uncooperative parties (Hardin Reference Hardin1982, 153–154). Norms of difference and exclusion govern large-number relationships of identification of a whole community and exclusion of others from it. They are reinforced through the mechanism implicit in their functional structure, not directly through the iterated incentive offered by the fellow group members in a large-number prisoner's dilemma. The explanation of large-number norms in general cannot turn on a rational regulation of the problem of collective action in enforcing the norm, although there may be some cases that can be regulated through somewhat unstable conventions (Hardin Reference Hardin1982, 173–205).

Norms that govern large-number interactions are therefore somewhat more difficult than dyadic norms for, say, promise keeping. If one of these governs a large-number prisoner's dilemma, then it would seem to be subject to free riding that would ruin the effectiveness of the norm. What we need is a system for motivating cooperative behavior in such an interaction. Many larger-number norms can readily be fitted to rational-choice principles through a simple feedback model of functional explanation. For example, suppose we wish to assert that a norm of loyalty to a group exists. We might crudely suppose that it exists somehow because it supports, say, intragroup cooperation. That is to say, the norm is functional for cooperation. Now we can explain the maintenance of the norm (but maybe not its origin) by the paradigm of functional explanation.

An institution or a behavioral pattern X is explained by its function F for group G if and only if:

1. 1. F is an effect of X;

2. 2. F is beneficial for G;

3. 3. F maintains X by a causal feedback loop passing through G.

The pattern X here is following the norm of basic loyalty to a group; its function F is social cooperation (or other related concerns) for the group.

1. 1. Now cooperation (F) is an effect of following the norm (X), that is, cooperation is built up of actions of keeping to the norm.

2. 2. Cooperation (F) is good for—serves the interests of—the members of the group (G).

3. 3. Their cooperation (F) maintains their loyalty (X) by a feedback loop passing through the members of the society (G).

Let us elaborate on step 2 in this explanation. Their cooperation is good for the members of the group. It would be wrong to say that the norm of group loyalty is a norm that serves our interests generally or that it is inherently good. This functional account works for many groups, including morally awful groups such as the Ku Klux Klan or a street gang. It also works for church groups, academic seminars, and so forth.

A large fraction of all standard large-number norms take this form. But that means that they are based in communal relationships. They cannot be readily studied in contexts that abstract from behavior in an ongoing set of relationships. Laboratory games often stipulate conditions that set aside all of this standard context that actually defines and enforces standard norms on the functional, rational-choice account. Gamers might suppose that players invent norms in context. Bicchieri generally supposes that the issues and contexts of her games evoke already extant norms, such as for fairness, to develop a more specific norm. Does this, as Bicchieri says, turn a large-number prisoner's dilemma into a coordination interaction? If I could cheat—be a snitch, for example—it is possibly a prisoner's dilemma. What would make it a coordination game?

3. Motivation

Bicchieri's condition b for a social norm to exist in situations of type S is: “Normative expectations: i believes that a sufficiently large subset of P (the relevant population) expects i to conform to [norm] R in situations of type S.” This seems to make i actually “prefer” to conform to the norm (Reference Bicchieri2006, 11). This preference is not well argued. If we invoke Bicchieri's second version of this condition b′—the version that includes sanctioning for noncompliance—we are left with the need to explain the endogenous and presumably spontaneous sanctioning by others in the group or society. Condition b′ is stronger: “Normative expectations: i believes that a sufficiently large subset of P (the relevant population) expects i to conform to R in situations of type S, prefers i to conform, and may sanction behavior.” In this version of the condition, we may have motivation to conform from sanctions, although there is no account of why anyone would sanction someone not complying with a norm.

Note an implication of condition b, which does not include sanctions. Here i must be normatively committed to following the norm. The claim seems to be that if there is a norm, people are normatively committed to following it. Either this is a trick of our psychology, or motivation, which one might have thought we must explain, has been folded into the definition of a norm. Others’ expectations of what I should do will motivate me to do what they expect of me. I think this claim is true for a class of expectations that Bicchieri does not count as norms. Simple coordinations, such as the driving convention of driving to the right (or the alternative convention of driving to the left) are not instances of Bicchieri's “situations of type S, because S can be represented as a mixed-motive game” (Reference Bicchieri2006, 11).Footnote ¹ In coordination interactions, we have strong internal motivation (of acting in our own interest) to follow the convention in place where we are. Bicchieri calls such conventions “descriptive norms.”

It is reasonable for Bicchieri to differentiate descriptive from her proper norms because the latter help to overcome one's interest in doing other than what the norm calls for. Without the sanctions of condition b′, there is no such easy internal motivation as there is for following descriptive or coordination norms. I always urgently want to follow the driving convention and I have no competing desire to cheat on that convention. But in many contexts I might have a competing desire to violate the norm of truth telling or promise keeping. Hence, we need an account of why the desire to follow the relevant norm overcomes any competing desire to violate it. Bicchieri says, if condition b is met, then “as a result, i prefers to conform to R” (emphasis added). Why? Without a rigorous answer to this question, the treatment of norms here is partially a throwback to the oldest—religious and moral—literature on norms, in which normative commitments are simply stipulated.

4. Concluding Remarks

The principal difficulty of a gaming account of norms is to show how the account explains motivations of individuals to follow the norms. Bicchieri often explains compliance with norms as the result of psychological anomalies or mistaken perceptions about the world and others’ likely intentions (Reference Bicchieri2006, 182ff.). It is the first concern of a rational-choice account to motivate individuals, but the rational-choice model of motivation is relatively simplistic. An advantage of this simplicity is that we do not need ad hoc stories of why a particular psychological anomaly is at work in some behavior. But maybe the potentiality for ad hoc storytelling is an implicit risk for an adequate account.